Early detection of ovarian cancer remains a challenge.

Pathologic changes within an organ might

be reflected in proteomic patterns in serum or plasma. The objective of this study was to identify
new plasma biomarkers in ovarian cancer patients using mass spectrometry (MS) protein
profiling and artificial intelligence. The study included 35 women with ovarian cancer and 30
age-matched disease-free controls.
For plasma protein signature analysis, the protein chip array surface-enhanced laser desorption/
ionization (SELDI) analysis was performed. (Plasma proteomic pattern as biomarkers
for ovarian cancer)

Of the gynecological malignancies, ovarian cancer has the highest mortality rate. Hidden deep in
the pelvis, ovarian cancer lacks early symptoms and signs, often presents at advanced stages, and
is associated with a low 5-year survival of 35%(15). More women die from ovarian cancer than
from all other gynecological malignancies combined. By contrast, the 5-year survival for patients
with stage I ovarian cancer exceeds 90%, and most patients are cured of their disease by surgery
alone(35). Therefore, increasing the number of women diagnosed with stage I disease should
have a direct effect on the mortality and economics of this cancer. The identification of cancer
biomarkers provides the possibility for early detection, better monitoring of tumor progression,
and even targeting therapy(17).
Traditional strategies for cancer biomarker identification used tumor cells to immunize animals
and screen for antibodies that could efficiently recognize the antigen(8).This approach is limited
by its high cost and labor intensity but has produced the best-known marker, CA125, approved
for ovarian cancer monitoring
( 2006 SELDI ovary)

Cytoreductive surgery and combination chemotherapy have improved outcomes in ovarian

cancer over the last three decades. However,attempts to improve survival by adding additional
cytotoxic drugs to carboplatin and paclitaxel have not succeeded [1]. Within the past few years,
substantial hope has been placed in therapy that targets specific abnormalities in ovarian cancer
cells. To personalize care by matching targeted therapy to specific abnormalities, an in depth
analysis of different subtypes of ovarian cancer will be required. More than 300 high-grade
serous epithelial ovarian cancers have been characterized copy number abnormalities, RNA
expression, microRNA expression and promoter methylation [2]. A similar study has not yet
been performed for the less common low-grade serous carcinomas (LGSC). The Gynecologic
Cancer InterGroup (GCIG) Consensus Conference has, however, recently affirmed the need to
develop specific treatment options/trials for the histological subtype of
LGSC [3]. Epithelial ovarian cancer (EOC) is a heterogenous disease which includes
five main histologic subtypesserous,mucinous, endometrioid, clear cell, and transitional cell
carcinomas [59]. Epithelial ovarian cancers can also be classified with regard to invasiveness
and degree of differentiation. Tumors of low malignant potential (LMP) generally
lack stromal invasion. Invasive carcinomas have been considered lowor high-grade [10].
Systems for grading serous carcinomas have not been universally accepted and have used
inconsistent criteria and categories, creating confusion regarding the optimal definitions [1215].
As we approached the 21st century, clearer pathologic criteria for low- and high-grade lesions
were warranted for planning therapy for individual patients and in performing clinical trials.

Traditionally, serous carcinomas have been classified as high-, intermediate, or low-grade [11].
After over a decade of experience, a two-tier system with clear morphologic criteria was reported
and subsequently validated, and inter- and intra-observer reproducibility was
demonstrated (Table 1) [16,17].
Evidence to date indicates that there is a close relationship between LGSC and serous LMP
tumor. Some 60% of LGSC are associated with serous LMP tumors [16], but the LGSC
component has prognostic significance, and it is essential that LGSC be distinguished from LMP
tumors. In addition, when stage IIIV serous LMP tumors recur, the vast majority is LGSC. In
one study, when relapses were observed following an initial diagnosis of an LMP tumor, 73%
contained LGSC elements[23]. Different histologic patterns of LGSC have recently been
stressed considering the difficulty of the diagnosis [24].
The classic distinction between serous LMP tumors and LGSC focuses on the degree of
infiltrative growth. The World Health Organization(WHO) definition distinguishes both entities
by the presence or absence of obvious destructive stromal invasion [25]. The presence
of microinvasion in serous LMP tumors with a maximal invasive focus size of 10 mm2 does not
affect prognosis in most studies [26,27], but it has in others [28,29]. Critics of this view suggest
that there is compelling evidence that microinvasion could represent an early stage in the
pathogenesis of LGSC [30].
Activation of the MAP kinase pathway is observed in up to 80% of LGCS and in 78% of its
putative precursor, LMP tumors [46]summarizes the published data regarding BRAF, KRAS,
p53, EGFR and HER2mutational analyses in LGSC. All of these studies involvemutually
exclusive BRAF mutations (235%) or KRAS mutations (054%), or HER2/neu amplification
(8%) [4758]. Only 8% of LGSC harbor p53 mutations compared to a frequency of 96% in highgrade serous carcinomas, as described in the recent Cancer Genome Atlas report [52,59].
Wong et al. reported a much lower frequency of BRAF mutations (2%) in advanced LGSC than
previously observed [58]. This finding questions the role of this mutation in early LGSC
pathogenesis and is not consistent with the apparent dependency on one or a few
genes for the maintenance of the malignant phenotype termed oncogenic
addiction by Weinstein [60]. In most series, however, including the study of Wong et al., the
frequency of KRAS mutations is 20% or higher. In this small series of 43 cases, LGSC without
BRAF and KRAS mutations had an overall survival of 47.3 months (95% CI, 22.12 to
72.5) compared to 78 months in the presence of either mutation (p = 0.28) [58]. This finding
suggests that KRAS/BRAF mutations may behave as a favorable prognostic biomarker. Since a
much higher frequency of BRAF mutations occurs in serous LMP tumors compared
to LGSC, it is possible that a fraction of LGSC do not arise from serous LMP tumors and are
associated with a BRAF mutation [58]. Additionally, this study was the first to suggest that the
presence of a BRAF mutation in a serous LMP tumor may actually be protective against
the development of a subsequent LGSC.
In a more recent report, Grisham et al. confirmed a number of the findings of the MD Anderson
group [61]. Of 19 patients with LGSC,only 1 (5%) had a BRAF mutation, whereas 16% had a
KRAS mutation.
They also agreed with theory of Wong et al. that a BRAF mutation in a serous LMP tumor may
somehow identify those patients who are unlikely to progress to a LGSC.
Very little study has focused on the PI3K/AKT pathway in LGSC.
Wang et al. performed mutational analysis for PIK3CA mutations in

90 serous carcinomas of the ovary and found three mutations, one of which was found in a
LGSC [62]. Clearly, further study of the PI3K/AKT pathway in relationship to LGSC is
warranted, particularly given the hypothesis that dual blockade of both the MAP kinase and
PI3K/AKT pathways may be necessary to achieve meaningful clinical benefit.
In a recent report, investigators performed exome sequencing analysis of 8 affinity-purified
LGSC [63]. A total of 70 somatic mutations in 64 genes were identified. However, except for
KRAS and BRAF mutations, none of the mutations were validated in an independent validation
set of LGSC and serous LMP tumors.
Compared to high-grade ovarian cancers, serous carcinoma most commonly, LGSC tends to
occur at a younger age and is associated with prolonged overall survival. Gershenson et al. first
reported their experience with 112 women with stage IIIV LGSC [90]. The median age was 43
years, and the median overall survival was 82 months. The other major finding was the apparent
relative insensitivity to chemotherapy as reflected in the low rate of patients who were clinically
disease-free at the completion of primary platinum-based chemotherapy (52%) and the low
negative second-look rate in the 39 patients with sufficient information (5%). Subsequently, this
relative chemoresistance was also observed in the analysis of patients undergoing neoadjuvant
chemotherapy and in patients treated for recurrent LGSC [91,92]. Schmeler et al. found that only
one of 24 evaluable patients who underwent platinum-based neoadjuvant chemotherapy had an
objective response based on imaging studies. On the other hand, of the 20 patients with sufficient
serum CA 125 data, 50% had a >50% reduction in levels after neoadjuvant chemotherapy [91].
Similarly, in patients with recurrent LGSC, the overall response rate to conventional
chemotherapy in 108 evaluable patient-regimens was only 3.7% [92]. On the other hand, the rate
of stable disease in both studies was over 60%. Some have misinterpreted these findings and
concluded that chemotherapy has no place in the armamentarium of options for women
with LGSC. In fact, this is not our conclusion. Rather, one simply needs to understand the
relative efficacy compared to that observed in patients with high-grade serous carcinomas and
the clinical benefit demonstrated by the high stable disease rate. In addition, the discordance
between serum CA 125 response and conventional computed tomography findings highlights the
need for a search for better imaging modalities for this rare ovarian subtype. Nevertheless, this
body of work strongly underscores the need for more effective treatments.
Currently, standard primary treatment for women with stage IIIV LGSC remains cytoreductive
surgery followed by platinum-taxane chemotherapy or, in patients with extensive tumor burden
or significant co-morbidities, neoadjuvant chemotherapy and interval debulking surgery
followed by further chemotherapy. In our experience, stage I LGSC is very uncommon, and there
is scant information regarding its clinical course. However, based on limited data, observation
may be a reasonable option. Clearly, more information is needed to formulate an
evidence-based approach for early-stage disease.
For women with recurrent LGSC, secondary cytoreductive surgery may be beneficial in selected
patients [23,93]. For systemic therapy, conventional chemotherapy also remains an important
option. For patients with platinum-sensitive recurrence, options include re-treatment
with paclitaxel/carboplatin, gemcitabine/carboplatin, docetaxel/ carboplatin, or pegylated
liposomal doxorubicin/carboplatin. For patients with platinum-resistant disease, options include
pegylated liposomal doxorubicin, topotecan, gemcitabine, docetaxel, orweekly
paclitaxel, among others.

Hormonal therapy has also been a therapeutic option for women with LGSC for several years,
although there is scant information in the literature. Wong et al. found that estrogen and
progesterone receptor positivity in LGSC is 58% and 43% compared to 27% and 17%
in high-grade serous carcinoma [73]. Unfortunately, most reports of hormonal therapy for
ovarian cancer include relatively small numbers of patients andmultiple histologic subtypes.
Although we lack prospective trials specifically regarding LGSC, some approaches
showpromising data. (CO low grade serous carcinoma 2013).
Ovarian cancer is immunogenic. The ability of the immune system to recognize ovarian cancer is
associated with improved prognosis.
The form of immunity associated with this improved prognosis is known; T cell infiltrates in
ovarian cancers are shown to be associated with improved prognosis in a number of studies. The
infiltration of T cells has been observed in ovarian cancers since 1982 [1]. The full
prognostic significance of T-cell infiltration in ovarian cancers, that it rivaled optimal surgical
cytoreduction, was subsequently reported by Zhang et al in 2003 [2]. The presence of
intratumoral T cells was an independent prognostic factor for PFS and OS by multivariate
analysis.
These findings have been validated in several subsequent studies, and point to the specific
importance of cytotoxic CD8+ T-cells [39].
Regulatory T cells, another subset of T cells that can modulate immune responses and maintain
tolerance to self-antigen, have been shown to predict poor patient survival in ovarian cancer
[6,10].
The natural pathobiology of ovarian cancer also allows opportunities for therapeutic vaccines to
be applied. Although over 60% of women diagnosed with ovarian cancer will have distant
metastases according to the most recent NCI SEER data, response rates to initial chemotherapy
and cytoreductive surgery can be as high as 85% [11]. Unfortunately, despite these initial
responses, over two thirds of patients will recur and even in patients who achieve complete
remissions, maintaining these has proven elusive [12]. Despite advances in therapies, cure rates
have changed little and most patients can expect a relapsing and remitting clinical course of
progressive resistance to chemotherapies. However, periods of remission could allow vaccines
the necessary time in patients with low disease burdens to induce an effective antitumor response
to prolong remissions and prevent recurrences.
Finally, we have already seen the ability of novel therapies, modulating T-cells, demonstrate
responses in ovarian cancer patients.
Cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) engagement of costimulatory molecules
can result in the arrest of T cell responses and impaired antitumor response. When antibody
blockage of CTLA-4 was used in heavily pretreated ovarian cancer patients, who had received
multiple lines of chemotherapy, four of nine patients were able to achieve stable disease, by CA125 and radiographic criteria, without significant toxicities [13,14]. Programmed death 1 (PD-1)
is another T-cell coinhibitory receptor. Antibody blockage of its ligand, PD-L1, has been studied
in patients with selected advanced cancers, including ovarian cancer. Patients with advanced or
metastatic disease having
failed at least one line of chemotherapy were treated with anti-PD-L1 antibody in an effort to
block inhibitory signals on effector T cells. 18%
of ovarian cancer patients (n = 17) were able to achieve stable disease for at least 6 months [15].
The success of these recent methods of T cell

modulation in an antigen non-specific fashion, in pretreated patients, suggests that vaccines

capable of generating focused immune responses specifically targeting tumor antigens may be
even more effective.
What is the track record of ovarian cancer vaccine therapy?
An ideal antigen for an ovarian cancer therapeutic vaccine would be solely expressed on ovarian
cancer cells, be highly immunogenic with a bias toward a cytotoxic antitumor response, and be
able to be carried or expressed using the chosen vaccine platform. Additionally,
the target should be biologically necessary in maintaining the malignant phenotype so that tumor
cells cannot escape immune targeting through loss of expression. These would be considered
tumor specific antigens. Few, if any, candidate antigens will meet all these criteria. HPV
oncoproteins E6 and E7 in cervical cancer are one of the few examples of this, viral proteins that
are also required for the malignant phenotype. For most malignancies, vaccine targets
represent tumor-associated antigens, which are over expressed in tumor cells, but also are present
in lower quantities in normal cells [16]. Because they are self-antigens, they are inherently less
immunogenic. Candidate antigens being evaluated in ovarian cancer generally
fall into this classification. A number of candidate antigens including HER2/neu, p53, CA125,
MUC1, CEA, folate receptor alpha, cancer testis antigens like NY-ESO-1 and insulin growth
factor binding proteins have all been proposed as potential vaccine targets in ovarian cancer
due to their reported immunogenicity [1727]. Therapeutic cancer vaccines have been evaluated
using a number of platforms including peptides/protein or DNA in combination with
adjuvant, anti-idiotype vaccines, recombinant viruses or other microbes, tumor cells or tumor
cell lysates, or the delivery of activated dendritic cells to patients. A number of these strategies
are currently being studied for ovarian cancer and their advantages and limitations
can be influenced by factors inherent to the specific platform (Table 1).
Peptide strategies are attractive because they allow the direct translation of an identified tumor
associated antigen into a vaccine and precise measurement of immune responses. Peptides of a
specific length and sequence can represent epitopes that may be presented on MHC molecules to
effector T cells. However, peptides and proteins have limited ability to elicit balanced and
durable CD4 and CD8 responses alone. Peptide and protein based vaccine platforms are usually
administered with an immune modulator or adjuvant because they are only weakly
immunogenic. These vaccines may only represent a portion of a tumor-associated antigen and
selection of epitopes may be limited by the diversity of HLA alleles in patients that are able to
recognize these epitopes. Long peptides incorporating both CD8+ and CD4+ epitopes have the
potential to be more efficiently presented to T cells and have been demonstrated in vaccination
targeting HPV E6 and E7 in cervical cancer [28]. This strategy has been reported in a phase 1
trial of 28 ovarian cancer patients using overlapping long peptides from a human tumor selfantigen, NY-ESO-1 with adjuvant. The vaccine was well tolerated and able to induce both
cellular, CD4+ and CD8+, and antibody responses in nearly all vaccinated patients when given
with a Poly-ICLC adjuvant [29].
While a peptide or protein strategy may be limited by the knowledge of and ability of a specific
patient's MHC molecules to present the selected amino acids sequences, it has the potential to
target multiple antigens. Additionally, downregulation of surface MHC class I is hypothesized to
be a strategy of immune evasion in a number of malignancies. A look at the feasibility of selected
peptides from candidate antigens: p53, SP17, survivin, WT1, and NY-ESO-1 to be
incorporated in a multiantigen vaccine was undertaken by Vermeij

and colleagues. In tumor samples from 270 primary ovarian cancer patients, 93.2%
overexpressed at least one of the candidate antigens.
Over 70% of patients overexpressed 2 or more of the candidate antigens.
The authors also found that expression of MHC class I was present in over 78% of ovarian
cancers tested. This combination of findings suggests that a vaccine directing a cellular immune
response against multiple target antigens may find some success in ovarian cancer [30].
Plasmid DNA vaccines have distinct advantages over synthetic peptides. DNA is more stable and
is easy to manufacture. It can be given in a smaller volume than peptide and may be better suited
to a multiantigen approach. Plasmid DNA can generate intracellular synthesis of the target
antigen allowing presentation via MHC class I and has been shown to be able to stably transfect
skin cells. It also is weakly immunogenic and requires the coadministration of an adjuvant, or the
use of a delivery vehicle such as a gene gun or liposomes. Additionally plasmids may be
modified to encode non-self antigens to enhance immunogenicity. This approach has been
explored in a number of malignancies, and is currently employed in phase I clinical trial at the
University ofWashington as an approach to vaccinate against IGFBP-2 for the treatment of
advanced or recurrent ovarian cancer in remission. Although the study is ongoing, preliminary
results regarding toxicity thus far are reassuring. There have been a total of 130 adverse events in
22 patients. 95% of adverse events to date were Grades 12 and 5% were Grades 35. The one
related Grade 3 event was decreased lymphocyte count, and not considered a dose limiting
toxicity. In preliminary analysis for immunogenicity, 40% of patients have demonstrated new
serum antibodies recognizing IGFBP-2 after completion of vaccinations (D. Cecil, personal
communication). Viral and microbial vectors have also been explored as vehicles for therapeutic
cancer vaccines. Exploiting their natural immunogenicity as pathogens obviates the need for
adjuvants and their genomes may be manipulated to express candidate vaccine targets. Selection
of an individual vector is crucial since each possesses specific characteristics in how it is
perceived by the immune system. Poxviral vectors, due to their large genome, can allow the
insertion of multiple genes, up to 10 kb, which may include the target antigen or antigens and
immunostimulatory factors. Poxviruses are double stranded DNA viruses that replicate in the
cytoplasm of infected cells, so no viral sequences are inserted in the host genome. In a trial
conducted at the National Cancer Institute, 26 patientswith metastatic breast or ovarian cancer
that had 3 ormore prior chemotherapy regimens were vaccinated with a recombinant poxviral
vaccine expressing tumor associated antigens,MUC-1 and CEA in addition to TRICOM, a set of
T cell stimulatory molecules including B7.1, ICAM-1, and LFA-3. (Therapeutic vaccines for
ovarian cancer)